Unnecessary /bin/lsblk executions in the get_mount_facts function of the setup module

[kilburn]

Issue Type:

• Bug Report

Ansible Version:

ansible 2.0.0.2
  config file = /Users/kilburn/Documents/Projects/tu/plataforma/ansible.cfg
  configured module search path = Default w/o overrides

ansible 2.1.0 (devel 517acb5773) last updated 2016/02/18 08:46:22 (GMT +200)
  lib/ansible/modules/core: (detached HEAD 8d126bd877) last updated 2016/02/18 08:46:32 (GMT +200)
  lib/ansible/modules/extras: (detached HEAD f6c5ed987f) last updated 2016/02/18 08:46:39 (GMT +200)
  config file = 
  configured module search path = Default w/o overrides

Ansible Configuration:

None relevant.

Environment:

Controller is OS X 10.10.5, the managed OS is Debian Jessie 8.3 (fully updated at the time of this writing).

Summary:

This is probably related to #10779. Function get_mount_facts times out on my test machine, throwing the following error:

test | FAILED! => {
    "changed": false, 
    "cmd": "/bin/lsblk -ln --output UUID /dev/dm-0", 
    "failed": true, 
    "msg": "Traceback (most recent call last):\n  File \"<stdin>\", line 1999, in run_command\n  File \"<stdin>\", line 2156, in _handle_timeout\nTimeoutError: Timer expired\n", 
    "rc": 257
}

The root cause is that, to setup multiple chroots, my test machine has many bind mounted paths. These paths appear in /etc/mtab, for instance:

/dev/dm-0 /srv/user1/usr/lib/x86_64-linux-gnu/ImageMagick-6.8.9 ext4 rw,relatime,errors=remount-ro,data=ordered 0 0
/dev/dm-0 /srv/user2/usr/lib/x86_64-linux-gnu/ImageMagick-6.8.9 ext4 rw,relatime,errors=remount-ro,data=ordered 0 0
/dev/dm-0 /srv/user3/usr/lib/x86_64-linux-gnu/ImageMagick-6.8.9 ext4 rw,relatime,errors=remount-ro,data=ordered 0 0
/dev/dm-0 /srv/user4/usr/lib/x86_64-linux-gnu/ImageMagick-6.8.9 ext4 rw,relatime,errors=remount-ro,data=ordered 0 0
/dev/dm-0 /srv/user5/usr/lib/x86_64-linux-gnu/ImageMagick-6.8.9 ext4 rw,relatime,errors=remount-ro,data=ordered 0 0
...

In the current implementation, get_mount_facts calls the lsblk binary to get the uuid of each device (first field of a line starting with /, such as those lines above). This means that lsblk is being called here many times to get the uuid of the same /dev/dm-0 device. That many calls make the function end up timing out.

Since the uuid of /dev/dm-0 is always going to be the same, all these calls could be avoided by caching the results of the first lsblk /dev/dm-0 execution.

Steps To Reproduce:

Setup a large number of bind-mounts on a test machine test. Run ansible -m setup test.

Expected Results:

You get the machine's facts.

Actual Results:

test | FAILED! => {
    "changed": false, 
    "cmd": "/bin/lsblk -ln --output UUID /dev/dm-0", 
    "failed": true, 
    "msg": "Traceback (most recent call last):\n  File \"<stdin>\", line 1999, in run_command\n  File \"<stdin>\", line 2156, in _handle_timeout\nTimeoutError: Timer expired\n", 
    "rc": 257
}